Personality Profiles of Victims of Intimate Partner Violence and Inmates: Contributions of the Personality Assessment Inventory and the Minnesota Multiphasic Personality Inventory-2-Restructured Form

Abstract

Although there is a growing body of research focused on the personality characteristics of victims and offenders, only a few studies have investigated both groups through robust and comprehensive measures of personality. The present study aimed to compare the PAI and MMPI-2-RF profiles between victims and offenders and investigate the influence of adverse childhood experiences (ACEs) on their results. Samples of 107 female victims (age: M = 42.71; SD = 11.25) and 154 male inmates (age: M = 36.51; SD = 12.72) were compared, and statistically significant differences were found on several PAI and MMPI-2-RF scales. While the victims tended to score higher on scales such as Anxiety, Stress, Somatic Complaints and Thought Dysfunction, the inmates scored higher on scales related to Antisocial Traits, Drug Problems, and Aggressiveness-Revised, among others. Both groups reported a large number of ACEs, and linear regression analyses revealed that ACEs predicted PAI and MMPI-2-RF scores. A discriminant analysis also found that specific ACEs accurately discriminate psychological characteristics between victim and offender groups. In conclusion, the PAI and the MMPI-2-RF provided valuable information on the characteristics of victims and inmates, contributing to a better understanding of the nature of victimization and crime perpetration.

1. Introduction

Forensic psychology can be understood as a field that applies psychological knowledge and research to subjects in the justice system and to legal settings (DeMatteo et al. 2016). As such, forensic psychologists are intimately familiar with different case proceedings, and can play an active role as experts in legal cases. Often, forensic psychologists conduct psychological assessments of either victims or offenders, measuring their cognitive abilities, personality, social desirability, malingering, and risk of recidivism, among others (Melton et al. 2018).

However, despite the growing number of instruments specifically designed to assess individuals in the legal system, the Personality Assessment Inventory (PAI; Morey 1991), the Multiphasic Personality Inventory 2 (MMPI-2; Butcher et al. 2001), and the Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF; Ben-Porath and Tellegen [2008] 2011) are the most frequently used tools by forensic psychologists (Meaux et al. 2022).

The PAI is an objective self-report designed to provide data on a person’s psychopathology, personality, and psychosocial environment, which are relevant constructs for diagnostic and clinical decision-making purposes (Morey and Boggs 2004). Its adequacy to forensic settings is widely recognized (e.g., Jackson and Hess 2007; Viljoen et al. 2010) due to its short administration times (i.e., 45 to 50 min), readability (i.e., at a fourth-grade level), and response validity scales (i.e., it provides scales that assess infrequent and inconsistent responses, as well as negative or positive impression management). These characteristics are important in settings in which a quick administration is needed, where people tend to have lower educational levels, and where malingering or response management are highly prevalent (Gardner and Boccaccini 2017; Morey 2003; Morey and Quigley 2002). On the other hand, the MMPI-2-RF is the 338-item restructured form of the MMPI-2, which includes 51 new and revised validity and substantive scales (Ben-Porath and Tellegen [2008] 2011; Friedman et al. 2015). Its utility in forensic settings has been acknowledged by researchers due to its psychometric properties and validity scales. Indeed, the validity scales are particularly useful for the detection of malingering in forensic populations (e.g., Sellbom et al. 2010).

Historically, the psychological assessment of victims has focused primarily on female victims of Intimate Partner Violence (IPV), which is a consequence of the fact that women, when compared to men, remain particularly susceptible to physical abuse, rape, multiple and simultaneous types of abuse, as well as to death at the hands of their partners (Black et al. 2011; Violence Policy Center 2018), even if it has been recognized that men may also be victims (e.g., Breiding 2014). In this way, research has shown that IPV victims are more likely to develop mental health issues such as post-traumatic stress disorder (PTSD) (Johnson and Zlotnick 2009; Mechanic et al. 2008) and major depressive disorder (Mechanic et al. 2008). Although no specific personality profiles have been presented to explain why people become victims, some studies have revealed that women who have experienced IPV do tend to score higher in measures that assess schizoid, avoidant, self-destructive, schizotypal, borderline, and paranoid personality traits (Pereira et al. 2020; Pico-Alfonso et al. 2008).

Moreover, studies dedicated to the psychological assessment of battered women have found that past adverse childhood experiences (ACEs) may increase their involvement in abusive relationships in adulthood, serving as a potential risk factor for victimization (e.g., Thulin et al. 2021). Indeed, IPV victims tend to score higher on specific PAI scales, namely on the Negative Impression (NIM) scale, on the Anxiety-Related Disorder (ARD) scale and subscales (i.e., Obsessive–Compulsive—ARD-O, Phobias—ARD-P, and Traumatic Stress—ARD-T), on the Paranoia (PAR) scale and subscales (i.e., Persecution—PAR-P and Hypervigilance—PAR-H), and on the Borderline Features (BOR) scale and subscales (i.e., Affective Instability—BOR-A and Negative Relationships—BOR-N) (e.g., Cherepon 1994; Drury et al. 2009; McDevitt-Murphy et al. 2005). McDevitt-Murphy et al. (2005) also pointed out that IPV victims also score higher on the Anxiety (ANX), ARD, NIM, PAR, Somatization (SOM), Schizophrenia (SCZ), and Nonsupport (NON) scales, while scoring significantly low in Treatment Rejection (RXR).

In comparison to studies conducted with victims, research focused on offenders has been more frequent, in part due to the wide number of offender typologies (e.g., sexual offenders, IPV offenders, homicide offenders, among others). Furthermore, prison inmates are of particular interest to researchers and forensic psychologists because they tend to manifest higher levels of psychopathology (Gemeda 2013) related to the conditions prior to incarceration (e.g., ACEs, mental health, substance abuse) (e.g., Crick et al. 2023; Kazeem 2020) and during incarceration (e.g., lack of privacy, constant feelings of vigilance, constant interaction with other inmates, sentence duration) (e.g., Abilleira and Rodicio-García 2020). In terms of specific personality profiles of offenders, studies have pointed to a high prevalence of antisocial and psychopathic personality traits (e.g., Cunha et al. 2021; Robertson et al. 2020) or neuroticism (e.g., Dorling et al. 2024), but most recognize the need for further studies on the personality of offenders. To try to answer these questions, Cunha and Gonçalves (2013), through cluster analyses, identified three subtypes of batterers based on psychopathology and violence severity/frequency, namely non-pathological (i.e., with no significant elevations in clinical dimensions, and fewer antisocial traits), disturbed (i.e., with more clinical symptoms of somatization, obsession–compulsion, interpersonal sensitivity, anxiety, and psychoticism), and antisocial/violent (i.e., with high antisocial features, affective facet of psychopathy, and drug problems).

ACEs have long been considered a significant predictor of criminal behavior, as theorized by models such as the General Aggression Model (GAM; Anderson and Bushman 2002), the General Strain Theory (GST; Agnew 1992), and other developmental and life-course explanations of offending, such as the interactional theory (Thornberry and Krohn 2005, 2019). However, to date, no theory has sufficiently explained the way ACEs influence delinquent and criminal behaviors.

Studies that have relied on the PAI to assess inmate populations have found high scores on the Alcohol Problems (ALC), Drug Problems (DRG), Mania (MAN), PAR, Aggression (AGG), Antisocial Features (ANT), and BOR scales (e.g., Boccaccini et al. 2010, 2013; Burneo-Garcés and Pérez-García 2018; Abilleira and Rodicio-García 2020; Battaglia et al. 2021). Therefore, it is possible to identify some common score increases between offender and victim populations, particularly with respect to the PAR and BOR scales. Higher scores on these specific scales may be ascribable to the fact that both offending and victimization experiences frequently lead to paranoid thinking, characterological suspiciousness, or frank persecutory delusions, often associated with paranoid psychosis (e.g., Coid et al. 2016; Shah et al. 2018), and that both victims and offenders are likely to have severe personality disorders, such as Borderline Personality Disorder (e.g., Guzmán et al. 2024; Sansone et al. 2012). Nevertheless, more studies are needed to analyze victims’ and offenders’ profiles to investigate the hypothesis of a profile overlap and the influence of ACEs in both groups.

Studies that compare offenders and victims have found that psychological (e.g., levels of anxiety and depression, self-esteem, self-control, dominance) and trajectory characteristics (e.g., school commitment, parental monitoring) can effectively predict offender and victim group membership. For example, lower school commitment, lower parental monitoring, and lower self-control appear to significantly distinguish group membership for both offenders and victims (Jennings et al. 2010). Moreover, anxiety, depression, and low self-control are predictive of victim–offender overlap, while dominance tends to be solely associated with offending (van Gelder et al. 2015).

It is also important to recognize that in some cases, offenders have also been victims in the past, and these victimization experiences are a risk factor for perpetrating offenses in the future (e.g., Lauritsen and Laub 2007). For example, Tillyer and Wright (2014) found that many of the admitted offenders had been victims in the past, leading to an overlap of characteristics between the offenders and the victims (e.g., the feeling of being socially isolated, having a negative temperament, experiencing substance abuse problems, among others).

Studies with the MMPI-2-RF have also been conducted with offenders, but research on victim profiles is still lacking. Studies with offenders frequently focus either on malingering and defensiveness detection (e.g., Jones 2023; Steffan et al. 2010; Tarescavage et al. 2014; Wall et al. 2015) or on risk assessment. Previous findings revealed that the MMPI-2-RF can accurately predict violence, suicide, recidivism, and probation violation (Glassmire et al. 2016; Grossi et al. 2015; Koh and Kim 2023; Tarescavage et al. 2014).

On the other hand, studies with prison inmates have been mostly focused on sex offenders and the categorization of types of sex offending (e.g., Campanini 2020; VanSlyke 2018). One of the few exceptions was the study developed by Sellbom (2014), where inmates were organized according to five subtypes of psychological and psychopathological dimensions, such as (i) low levels/absence of externalization, antisocial behavior, low levels of negative emotionality, and thought dysfunction; (ii) egocentric/aggressive inmates with elevations in persecutory ideation, aberrant experiences, high levels of externalizing problems, ruthless aggression, and egocentric interpersonal style; (iii) psychoticism with high levels of thought dysfunction, aberrant experiences, mania, impulsive nonconformity, and experience seeking; (iv) internalizing with emotional dysfunction, low positive emotionality, introversion, and high negative emotionality; and (v) antisocial/disinhibited, which included severe levels of antisociality and impulsivity.

Campanini (2020) found that contact sex offenders scored higher than noncontact sex offenders on Thought Dysfunction (THD) and Behavioral/Externalizing Dysfunction, whereas Grossi et al. (2015) found that THD, Aberrant Experiences (RC8), Juvenile Conduct Problems (JCP), and Psychoticism-revised (PSYC-r) were associated with aggression in pre-trial defendants.

Overall, studies that have applied the PAI and the MMPI-2-RF simultaneously to forensic samples are scarce, and the existing studies have either focused solely on one of these groups (i.e., either victims or offenders) or on different types of crime perpetration (e.g., sexual abuse). In this context, two robust and psychometrically sound measures were used (i.e., the PAI and the MMPI-2-RF) to assess both the victim and the violent offender samples at once.

We investigated whether the victim and offender groups would differ in some personality traits. We also investigated whether some overlap would be found between the groups, particularly with regard to social isolation and negative affect. ACEs were expected to be found in both samples.

Aims

The present study aimed to compare the PAI and MMPI-2RF scores between victims and offenders and analyze the influence of ACEs in both groups. The administration of two comprehensive and robust measures of personality to both victimized and offending samples is relevant in that it can fill in the gaps regarding what we know about the personality styles of apparently contrasting samples and the way they may, or may not, overlap. Furthermore, by providing different scales and subscales from one another, these inventories complement each other regarding the domains assessed, thereby leading to a more complete and thorough understanding of the psychological functioning of both victims and offenders. This, in turn, can be helpful for mental health technicians who work firsthand with forensic samples by providing them with a wide array of information relevant to clinical diagnosis, treatment planning, and screening.

To our knowledge, no studies have used both the PAI and the MMPI-2-RF to assess victims and offenders. Studies of this nature (where different, but similar, measures are used and results are compared) could contribute to the validity of these measures in a relevant way, particularly their convergent and predictive validity. Therefore, the present study also aimed to investigate the convergent validity of the two questionnaires (the PAI and the MMPI-2-RF).

2. Methods

2.1. Study Design and Participants

This study employed a quantitative, cohort (i.e., non-experimental), and transversal design. The participants were 107 female IPV victims and 154 adult male prison inmates (see

Table 1. Demographic/psychosocial data and information on ACEs of the sample.

	Victims (n = 107)			Inmates (n = 154)
	%	M (SD)	Range	%	M (SD)	Range
Age (years)		42.71 (11.25)	20–73		36.51 (12.72)	18–74
Education
1st to 4th grades	9.3			1.9
5th to 6th grades	14.0			21.4
7th to 9th grades	21.5			36.4
10th to 12th grades	35.5			32.5
University	19.6			6.5
Type of crime
Homicide				42.2
Theft				27.3
Physical assault				26.0
Sexual offenses				20.8
Crimes against personal freedom				11.7
Road crimes				10.39
Number of offenses
Single offense				33.8
Multiple offenses				65.6
Recidivism
Yes				26.2
No				73.8
Years of incarceration					5.99 (6.69)	0–32
Adverse Childhood Experiences (ACEs)
Total of ACEs a		4.59 (4.18)	0–17		3.77 (3.93)	0–15
At least one ACE	81.3			79.9
At least four ACEs	49.5			42.9
Being insulted/humiliated	37.4			23.4
Fear of being physically hurt	43.0			27.9
Pulled/grabbed or thrown objects at	28.0			19.6
Hurt, leaving marks	37.4			31.4
Touched in a sexual manner	15.9			9.2
Sexually abused	10.3			22.2
Family did not love or support	36.4			28.8
Family did not look out for one another	38.3			28.8
Not enough food, or dirty clothes	16.8			20.3
Parents too drunk or sick to take care of the children	18.7			19.0
Parental separation/divorce	28.3			38.6
Parent was pulled, grabbed, bitten, or hit with objects	19.6			21.6
Parent was kicked or hit	20.6			21.6
Parent was repeatedly hurt with or without a weapon	15.9			17.0
Lived with someone with drug/alcohol problems	41.1			36.6
Lived with someone who was depressed/had psychiatric problems or tried to commit suicide	30.8			20.9
Lived with people who were arrested	8.4			20.1

Note: ^a Total of ACEs = frequency of ACEs.

). The victim group was recruited from a private practice specializing in forensic evaluations, shelters for female victims of violence, and non-governmental Portuguese victim support associations. The female IPV victims were aged between 20 and 73 years (M = 42.71; SD = 11.25), with most having completed the 12th grade (33.6%). The inclusion criteria were having suffered from IPV and speaking/reading Portuguese fluently. The exclusion criterion was not understanding Portuguese.

The prison inmate group participants were incarcerated at multiple high-security prisons in Portugal. The inclusion criteria were being incarcerated and speaking/reading Portuguese fluently. The exclusion criterion was not understanding Portuguese. The offender group had a mean age of 36.51 years (SD = 12.72), and most had completed the 9th grade (35.7%). This group committed crimes of homicide (n = 65), sexual offenses (n = 32), physical assault (n = 40), crimes against personal freedom (n = 18), theft (n = 42), and road crimes (n = 16). With regard to the number of committed offenses, 33.8% (n = 52) reported a single offense, while 65.6% (n = 101) reported having perpetrated more than one crime. Finally, 26.2% (n = 51) of the inmates were recidivists, meaning they had a prior history of incarceration. The inmate group had a mean of 5.99 (SD = 6.69) years of incarceration at the time of this study.

2.2. Measures and Procedures

The participants filled out a protocol comprised of a sociodemographic questionnaire, a questionnaire assessing ACEs, the PAI, and the MMPI-2-RF. All the questionnaires and inventories were in Portuguese. All data were collected in-person. The present study was approved by Faculty of Psychology and Education Sciences of the University of Coimbra Ethics Board (CEDI/FPCEUC:78/R_5). All the participants read and signed an informed consent form prior to their participation. The participants did not receive compensation for their involvement in the study.

2.3. Adverse Childhood Experiences Scale

The participants answered a brief questionnaire with 17 items about ACEs based on the work of Felitti et al. (1998) (see Table 1). This questionnaire measures childhood exposure to psychological, physical, or sexual abuse, neglect, mental illness, domestic violence, divorce, and having a parent in prison. The items were presented using a two-point, dichotomous-type scale (“yes” or “no”).

Higher scores suggested that the respondent had gone through more types of childhood adverse experiences. The type and frequency of experiences were also collected.

2.4. Personality Assessment Inventory

The PAI is a self-report measure of personality and psychopathology designed to provide information on several variables that are relevant to treatment planning, implementation, and evaluation (Morey 2004). It is comprised of 344 items and 22 nonoverlapping scales, including four validity scales, 11 clinical scales, five treatment scales, and two interpersonal scales. Furthermore, nine of its clinical scales and one of its treatment scales contain conceptually derived subscales (Morey 1991). Response options are presented using a 4-point Likert scale (“false”, “slightly true, “quite true”, and “totally true”).

The Portuguese version of the PAI showed adequate psychometric properties (e.g., reliability (e.g., α ANX = 0.92; α SUI = 0.91; α ANT = 0.73; for all alphas, see Paulino et al. (2024))). All the PAI scores used in the present study were raw scores, wherein higher scores suggest higher symptomatology (e.g., higher levels on the Anxiety scale reveal higher levels of anxious symptomatology).

2.5. Minnesota Multiphasic Personality Inventory-2-Restructured Form

The MMPI-2-RF (Ben-Porath and Tellegen [2008] 2011) is a 338-item self-report inventory designed to assess psychopathology symptoms and maladaptive personality traits by presenting a psychometrically improved version of the MMPI-2. It is comprised of 51 scales, including nine Validity Scales (VS), three Higher-Order (H-O) scales, nine Restructured Clinical scales (RC), 23 Specific Problems (SP) scales, the Personality Psychopathology Five scales (PSY-5), and two Interest scales (Sellbom 2019).

The Portuguese version of the MMPI-2-RF (Novo et al. 2023) was used for the present study, which revealed adequate psychometric properties (e.g., reliability with 4 scales with alphas ≥ 0.90; for further data, see Novo et al. (2023)). For the present study, raw scores were considered, and higher scores represent higher levels of symptomatology.

2.6. Statistical Analysis

Statistical analysis was performed using IBM Statistical Package for the Social Sciences Statistics (SPSS) Version 29. Group differences were analyzed using an independent t-test. Levene’s test for equality of variances was analyzed. In cases where Levene’s test was >0.05, equal variances were not assumed, and a Welch t-test was used instead. Cohen’s d was calculated to determine the effect size of the difference between groups: large if exceeding 0.80, moderate if at 0.50, and small if less than 0.20 (Cohen 1988).

To determine the influence of ACEs on the PAI and MMPI-2-RF scores, linear regression analyses were conducted. Total variance (R²) of the regression model and standardized regression coefficients (β) for each independent variable were reported. Discriminant analyses were also performed to assess whether ACEs differentiated between victims and offenders. The results were interpreted based on discriminant functions, structure matrix coefficients (i.e., correlation of discriminator variables with the function), and group centroids (i.e., the mean discriminant score of the members of a group on a discriminant function) (Brown and Wicker 2000).

3. Results

3.1. Comparisons in the PAI and MMPI-2-RF Scores Between the Victims and the Offenders

The independent t-test revealed that the raw PAI scores of the victim group were statistically higher than in the inmate group in SOM, ANX, ARD, Depression (DEP), SCZ, BOR, Suicide Ideation (SUI), and Stress (STR) (see

Table 2. Mean PAI scores for the victim group and the inmate group.

Scale a	Victims		Inmates		t (259)	p	Cohen’s d
	M	SD	M	SD
Validity scales
ICN	5.96	2.54	7.32	3.55	−3.56	<0.001	0.44
INF	4.34	2.10	6.42	2.61	−7.08	<0.001	0.88
NIM	4.27	3.49	3.20	2.73	2.66	0.009	0.34
PIM	15.40	3.97	16.60	3.76	−2.47	0.014	0.31
Clinical scales
SOM	22.50	13.76	14.17	9.68	5.35	<0.001	0.70
ANX	33.98	12.58	22.54	10.66	7.84	<0.001	0.98
ARD	32.46	11.07	25.11	9.18	5.59	<0.001	0.72
DEP	28.41	12.35	20.40	9.79	5.55	<0.001	0.72
MAN	25.57	9.96	25.79	11.17	−0.16	0.873	0.02
PAR	32.21	10.03	30.60	9.19	1.32	0.188	0.17
SCZ	19.57	8.46	16.26	8.50	3.07	0.002	0.39
BOR	32.45	11.19	27.49	11.49	3.45	0.001	0.44
ANT	13.00	6.62	24.49	9.99	−11.07	<0.001	1.36
ALC	2.55	3.84	6.26	6.39	−5.82	<0.001	0.70
DRG	3.31	3.08	8.81	7.94	−7.72	<0.001	0.91
Treatment scales
AGG	15.45	8.50	18.84	10.31	−2.78	0.006	0.36
SUI	7.69	7.26	4.39	5.60	3.93	<0.001	0.51
STR	12.51	4.47	9.11	4.07	6.37	<0.001	0.80
NON	7.24	4.41	7.38	4.61	−0.25	0.800	0.03
RXR	12.72	3.85	13.74	4.14	−2.01	0.046	0.26
Interpersonal scales
DOM	20.42	5.16	22.18	5.38	−2.63	0.009	0.33
WRM	23.10	5.26	22.79	6.07	0.42	0.672	0.05

Note: d = 0.20: small effect size; d = 0.50: medium effect size; d = 0.80: large effect size (Cohen 1988); ICN = Inconsistency; INF = Infrequency; NIM = Negative Impression; PIM = Positive Impression; SOM = Somatic Complaints; ANX = Anxiety; ARD = Anxiety-Related Disorders; DEP = Depression; MAN = Mania; PAR = Paranoia; SCZ = Schizophrenia; BOR = Borderline Features; ANT = Antisocial Features; ALC = Alcohol Problems; DRG = Drug Problems; AGG = Aggression; SUI = Suicidal Ideation; STR = Stress; NON = Nonsupport; RXR = Treatment Rejection; DOM = Dominance; WRM = Warmth. ^a Bonferroni correction, p = 0.002.

). In contrast, the inmate group scored significantly higher on the Inconsistency (ICN), Infrequency (INF), ANT, ALC, and DRG scales. No significant differences between the groups were found on the remaining scales. The magnitude of the differences ranged from small (d = 0.02 in MAN) to large (d = 1.36 in ANT).

The findings from the MMPI-2-RF (

Table 3. Mean MMPI-2-RF scores for the victim group and the inmate group.

Scale a	Victims		Inmates		t (259)	p	Cohen’s d
	M	SD	M	SD
Validity scales
VRIN-r	10.92	3.07	5.52	3.04	13.97	<0.001	1.77
TRIN-r	10.42	3.43	10.96	3.04	−1.29	0.198	0.17
F-r	15.60	3.40	7.07	5.09	16.12	<0.001	1.97
Fp-r	12.25	2.14	3.88	2.67	26.80	<0.001	3.46
Fs	6.39	2.22	2.95	2.62	11.00	<0.001	1.42
FBS-r	16.06	2.22	9.19	4.36	16.54	<0.001	1.99
RBS	13.18	2.35	7.72	3.73	14.37	<0.001	1.75
L-r	6.84	1.44	6.75	2.14	0.41	0.682	0.05
K-r	8.25	1.68	6.68	2.69	5.72	<0.001	0.70
Higher-order scales
EID	19.54	2.92	13.83	7.54	8.45	<0.001	1.00
THD	13.28	2.78	6.17	3.99	16.87	<0.001	2.07
BXD	9.81	2.67	9.42	4.59	0.85	0.395	0.10
Restructured clinical scales
RCd	11.27	2.36	7.72	5.59	6.97	<0.001	0.83
RC1	14.37	2.02	6.38	5.27	16.95	<0.001	2.00
RC2	8.97	2.10	5.56	4.03	8.83	<0.001	1.06
RC3	5.62	2.02	9.09	3.53	−9.95	<0.001	1.21
RC4	10.24	2.25	7.73	3.96	6.43	<0.001	0.78
RC6	8.30	2.29	4.44	2.84	12.03	<0.001	1.50
RC7	9.42	2.93	7.84	5.06	3.14	0.002	0.38
RC8	8.99	2.25	4.46	3.28	13.12	<0.001	1.61
RC9	11.06	3.74	13.11	5.07	−3.74	<0.001	0.46
Specific problems scales
Somatic/Cognitive
MLS	4.67	1.26	3.13	2.11	7.29	<0.001	0.89
GIC	2.65	0.91	0.62	1.01	16.52	<0.001	2.11
HPC	3.03	0.98	1.38	1.51	10.56	<0.001	1.30
NUC	6.01	1.28	2.62	2.42	14.60	<0.001	1.75
COG	4.80	1.71	2.28	2.28	10.11	<0.001	1.25
Internalizing
SUI	1.87	0.99	0.48	1.06	10.67	<0.001	1.36
HLP	3.07	0.83	1.64	1.13	11.70	<0.001	1.44
SFD	1.24	0.91	0.99	1.32	1.79	0.074	0.22
NFC	3.60	1.57	3.61	2.15	−0.02	0.981	0.01
STW	1.87	0.95	3.25	1.62	−8.62	<0.001	1.04
AXY	2.74	0.99	1.47	1.38	8.58	<0.001	1.06
ANP	3.65	1.10	2.20	1.89	7.74	<0.001	0.94
BRF	4.46	1.40	2.17	1.43	12.78	<0.001	1.62
MSF	4.21	1.36	2.91	2.05	6.11	<0.001	0.75
Externalizing
JCP	2.90	1.06	2.62	1.93	1.49	0.136	0.18
SUB	3.89	1.27	1.47	1.49	13.98	<0.001	1.75
AGG	3.98	1.79	2.66	1.99	5.45	<0.001	0.70
ACT	2.64	1.42	3.31	1.94	−3.19	0.002	0.39
Interpersonal
FML	5.22	1.40	2.39	2.30	12.21	<0.001	1.49
IPP	4.85	1.72	3.07	2.36	6.97	<0.001	0.86
SAV	5.25	1.85	3.04	2.71	7.79	<0.001	0.95
SHY	3.32	1.22	2.54	1.92	3.99	<0.001	0.48
DSF	2.45	0.99	1.13	1.21	9.67	<0.001	1.19
Interest scales
AES	4.07	1.00	2.70	1.96	7.34	<0.001	0.88
MEC	4.82	1.81	4.48	2.30	1.34	0.181	0.16
PSY-5
AGG-r	9.18	2.16	10.11	3.38	−2.68	0.008	0.33
PSYC-r	13.11	3.11	5.64	4.01	16.82	<0.001	2.08
DISC-r	7.58	2.50	8.93	3.83	−3.42	0.001	0.42
NEGE-r	8.72	1.73	8.26	3.54	1.38	0.170	0.17
INTR-r	10.44	3.34	6.92	4.74	6.99	<0.001	0.86

Note: d = 0.20: small effect size; d = 0.50: medium effect size; d = 0.80: large effect size (Cohen 1988). VRIN-r = Variable Response Inconsistency; TRIN-r = True Response Inconsistency; F-r = Infrequent Responses; Fp-r = Infrequent Psychopathology Responses; Fs = Infrequent Somatic Responses; FBS-r = Symptom Validity; RBS = Response Bias; L-r = Uncommon Virtues; K-r = Adjustment Validity; EID = Emotional/Internalizing Dysfunction; THD = Thought Dysfunction; BXD = Behavioral/Externalizing Dysfunction; RCd = Demoralization; RC1 = Somatic Complaints; RC2 = Low Positive Emotions; RC3 = Cynicism; RC4 = Antisocial Behavior; RC6 = Ideas of Persecution; RC7 = Dysfunctional Negative Emotions; RC8 = Aberrant Experiences; RC9 = Hypomanic Activation; MLS = Malaise; GIC = Gastrointestinal Complaints; HPC = Head Pain Complaints; NUC = Neurological Complaints; COG = Cognitive Complaints; SUI = Suicidal/Death Ideation; HLP = Helplessness/Hopelessness; SFD = Self-Doubt; NFC = Inefficacy; STW = Stress/Worry; AXY = Anxiety; ANP = Anger Proneness; BRF = Behavior-Restricting Fears; MSF = Multiple Specific Fears; JCP = Juvenile Conduct Problems; SUB = Substance Abuse; AGG = Aggression; ACT = Activation; FML = Family Problems; IPP = Interpersonal Passivity; SAV = Social Avoidance; SHY = Shyness; DSF = Disaffiliativeness; AES = Aesthetic–Literary Interests; MEC = Mechanical–Physical Interests; AGG-r = Aggressiveness-Revised; PSYC-r = Psychoticism-Revised; DISC-r = Disconstraint-Revised; NEGE-r = Negative Emotionality/Neuroticism-Revised; INTR-r = Introversion/Low Positive Emotionality-Revised. ^a Bonferroni correction, p = 0.001.

) showed that the victim group scored statistically higher on most scales, except for Cynicism (RC3), Hypomanic Activation (RC9), Stress/Worry (STW), and Disconstraint-Revised (DISC-r), where the inmates scored higher. No statistically significant differences between the groups were found in True Response Inconsistency (TRIN-r), Uncommon Virtues (L-r), Behavioral/Externalizing Dysfunction (BXD), Dysfunctional Negative Emotions (RC7), Self-Doubt (SFD), Inefficacy (NFC), Juvenile Conduct Problems (JCP), Activation (ACT), Mechanical–Physical Interests (MEC), Aggressiveness-Revised (AGG-r), and Negative Emotionality/Neuroticism-Revised (NEGE-r). The magnitude of the differences ranged from small (d = 0.01 in NFC) to large (d = 3.46 in Infrequent Psychopathology Responses—Fp-r).

3.2. Adverse Childhood Experiences in Victims and Offenders

Data on the type and frequency of ACEs were used. The most common ACEs for the victims were having felt scared of being physically hurt (43.0%; n = 46) and living with someone with drug/alcohol problems (41.1%; n = 44), while for the offenders, these were having lived through parental separation/divorce (38.3%; n = 59) and having lived with someone with drug or alcohol problems (36.4%; n = 56).

Linear regression analyses were conducted to investigate the predictive effect of ACEs (i.e., total number of ACEs) on the PAI and MMPI-2-RF scores for the victims and the offenders (see

Table 4. Linear regression analysis for ACEs predicting the PAI scores.

	β	R2	p
Validity scales
ICN	0.104	0.011	0.098
INF	0.032	0.001	0.611
NIM	0.262	0.069	<0.001
PIM	−0.168	0.028	0.007
Clinical scales
SOM	0.153	0.023	0.014
ANX	0.196	0.038	0.002
ARD	0.170	0.029	0.006
DEP	0.184	0.034	0.003
MAN	0.141	0.020	0.024
PAR	0.202	0.041	0.001
SCZ	0.185	0.034	0.003
BOR	0.245	0.060	<0.001
ANT	0.156	0.024	0.012
ALC	0.048	0.002	0.441
DRG	0.204	0.042	0.001
Treatment scales
AGG	0.264	0.070	<0.001
SUI	0.196	0.038	0.002
STR	0.204	0.042	0.001
NON	0.241	0.058	<0.001
RXR	−0.144	0.021	0.020
Interpersonal scales
DOM	0.016	0.000	0.801
WRM	−0.129	0.017	0.040

Note: ICN = Inconsistency; INF = Infrequency; NIM = Negative Impression; PIM = Positive Impression; SOM = Somatic Complaints; ANX = Anxiety; ARD = Anxiety-Related Disorders; DEP = Depression; MAN = Mania; PAR = Paranoia; SCZ = Schizophrenia; BOR = Borderline Features; ANT = Antisocial Features; ALC = Alcohol Problems; DRG = Drug Problems; AGG = Aggression; SUI = Suicidal Ideation; STR = Stress; NON = Nonsupport; RXR = Treatment Rejection; DOM = Dominance; WRM = Warmth.

). The results suggested that the number of ACEs contributes significantly to the prediction of scores on most PAI scales (i.e., NIM, PIM, SOM, ANX, ARD, DEP, MAN, PAR, SCZ, BOR, ANT, DRG, AGG, SUI, STR, NON, RXR, and WRM). The highest R² value is 0.070, for AGG, which indicates that 7.0% of the variance in AGG scores was explained by the number of ACEs.

The MMPI-2-RF scales were also significantly influenced by ACEs (see

Table 5. Linear regression analysis summary for ACEs predicting the MMPI-2-RF scores.

	β	R2	p
Validity scales
VRIN-r	0.234	0.055	0.016
TRIN-r	0.298	0.089	0.002
F-r	0.016	0.000	0.875
Fp-r	−0.093	0.009	0.345
Fs	0.389	0.151	<0.001
FBS-r	0.100	0.010	0.310
RBS	0.281	0.079	0.003
L-r	0.063	0.004	0.523
K-r	0.088	0.008	0.372
Higher-order scales
EID	0.136	0.019	0.164
THD	0.229	0.052	0.018
BXD	0.216	0.047	0.026
Restructured clinical scales
RCd	0.126	0.016	0.197
RC1	−0.212	0.045	0.029
RC2	−0.094	0.009	0.337
RC3	−0.078	0.006	0.428
RC4	0.062	0.004	0.528
RC6	0.112	0.012	0.255
RC7	0.296	0.088	0.002
RC8	0.297	0.088	0.002
RC9	0.226	0.051	0.020
Specific problems scales
Somatic/Cognitive
MLS	−0.298	0.089	0.002
GIC	0.133	0.018	0.173
HPC	−0.032	0.001	0.743
NUC	−0.078	0.006	0.426
COG	−0.008	0.000	0.936
Internalizing
SUI	0.071	0.005	0.471
HLP	−0.067	0.004	0.497
SFD	−0.089	0.008	0.365
NFC	0.117	0.014	0.231
STW	−0.034	0.001	0.727
AXY	0.010	0.000	0.917
ANP	0.173	0.030	0.076
BRF	0.020	0.000	0.837
MSF	−0.018	0.000	0.855
Externalizing
JCP	0.028	0.001	0.779
SUB	−0.112	0.013	0.252
AGG	0.137	0.019	0.161
ACT	0.108	0.012	0.270
Interpersonal
FML	0.050	0.002	0.612
IPP	−0.189	0.036	0.053
SAV	−0.066	0.004	0.501
SHY	0.221	0.049	0.023
DSF	0.304	0.093	0.002
Interest scales
AES	0.068	0.005	0.491
MEC	0.201	0.040	0.039
PSY-5
AGG-r	0.171	0.029	0.080
PSYC-r	0.250	0.063	0.010
DISC-r	0.215	0.046	0.027
NEGE-r	0.036	0.001	0.711
INTR-r	−0.119	0.014	0.223

Note. VRIN-r = Variable Response Inconsistency; TRIN-r = True Response Inconsistency; F-r = Infrequent Responses; Fp-r = Infrequent Psychopathology Responses; Fs = Infrequent Somatic Responses; FBS-r = Symptom Validity; RBS = Response Bias; L-r = Uncommon Virtues; K-r = Adjustment Validity; EID = Emotional/Internalizing Dysfunction; THD = Thought Dysfunction; BXD = Behavioral/Externalizing Dysfunction; RCd = Demoralization; RC1 = Somatic Complaints; RC2 = Low Positive Emotions; RC3 = Cynicism; RC4 = Antisocial Behavior; RC6 = Ideas of Persecution; RC7 = Dysfunctional Negative Emotions; RC8 = Aberrant Experiences; RC9 = Hypomanic Activation; MLS = Malaise; GIC = Gastro-Intestinal Complaints; HPC = Head Pain Complaints; NUC = Neurological Complaints; COG = Cognitive Complaints; SUI = Suicidal/Death Ideation; HLP = Helplessness/Hopelessness; SFD = Self-Doubt; NFC = Inefficacy; STW = Stress/Worry; AXY = Anxiety; ANP = Anger Proneness; BRF = Behavior-Restricting Fears; MSF = Multiple Specific Fears; JCP = Juvenile Conduct Problems; SUB = Substance Abuse; AGG = Aggression; ACT = Activation; FML = Family Problems; IPP = Interpersonal Passivity; SAV = Social Avoidance; SHY = Shyness; DSF = Disaffiliativeness; AES = Aesthetic-Literary Interests; MEC = Mechanical-Physical Interests; AGG-r = Aggressiveness-Revised; PSYC-r = Psychoticism-Revised; DISC-r = Disconstraint-Revised; NEGE-r = Negative Emotionality/Neuroticism-Revised; INTR-r = Introversion/Low Positive Emotionality-Revised.

): VRIN-r, TRIN-r, Fs, RBS, THD, BXD, RC1, RC7, RC8, RC9, MLS, SHY, DSF, MEC, PSYC-r, and DISC-r. All the remaining scales were not significantly predicted by the number of ACEs. The amount of ACEs contributes most to predicting scores in Fs (β = 0.389, R² = 0.151).

In addition, a discriminant analysis was performed to determine ACEs’ (i.e., different types of ACEs) ability to differentiate between the offenders and the victims (see

Table 6. Classification results of the discriminant analysis—types of ACEs ^a.

		Predicted Group Membership
	Group	Victims	Offenders
Original (%)	Victims	88	19
	Offenders	10	144
Cross-validated (%)	Victims	82.2	17.8
	Offenders	6.5	93.5
Canonical correlation	Wilks’ lambda	df	p
0.704	0.505	17	<0.001

Note: ^a 88.9% of the originally grouped cases correctly classified.

). The discriminant function was significant with a Wilks’ lambda of 0.505 (p < 0.001), a canonical correlation of 0.704, and an eigenvalue of 0.982. Watching a parent being repeatedly hurt/threatened with a weapon led to the highest structural load (0.754), followed by witnessing a parent being kicked or hit with an object (0.621) and witnessing a parent being pulled, grabbed, bitten, or hit with objects (0.612). The lowest structural load belonged to having parents who were too drunk or sick to take care of the children (0.013). The mean centroid of the discriminant function for the victims was 1.186, while for the offenders it was −0.822. An overall hit rate of 88.9% was found.

4. Discussion

The growing interest in validating psychological assessment measures for forensic populations has been reflected in the number of studies dedicated to the administration of widely recognized personality tests, such as the PAI and the MMPI-2-RF to victim and inmate samples. These studies contribute to the characterization of these populations by comparing the PAI and MMPI-2-RF scores between victim and inmate populations and by investigating the influence of ACEs in both groups.

We found statistically significant differences between the inmates and the victims on several PAI scales (victims > inmates: SOM, ANX, ARD, DEP, SCZ, BOR, SUI, and STR). Higher scores on these scales are consistent with the studies that investigated PAI profiles in victims (e.g., Cherepon 1994; McDevitt-Murphy et al. 2005). Higher scores in SUI and STR were also observed in the studies that linked IPV to stress and suicidality (e.g., McLaughlin et al. 2012; Yim and Kofman 2019).

Inversely, the inmate population scored significantly higher on the ICN, INF, ANT, ALC, and DRG scales. These results were expected, considering that inmate populations are frequently described as more likely to distort responses, and therefore score higher on validity scales (e.g., ICN, INF; Reidy et al. 2016) and in ANT (e.g., Buffington-Vollum et al. 2002) and ALC (e.g., Burneo-Garcés and Pérez-García 2018).

Interestingly, no significant differences were found between the groups in NIM, PIM, MAN, PAR, AGG, NON, RXR, DOM, and WRM, although both groups scored highly in PAR. These findings suggest that paranoia may be a prevalent trait in both victims (e.g., Pereira et al. 2020) and inmates (e.g., Abilleira and Rodicio-García 2020), but also support the belief that offenders and victims may have personalistic characteristics in common.

In relation to the MMPI-2-RF, we found that the offenders scored higher in RC3, RC9, STW, and DISC-r. Higher scores on these scales were expected, considering that they measure suspicion, grandiose self-view, sensation-seeking, risk-taking, poor impulse control, and anxiety (Friedman et al. 2015), all of which are commonly associated with antisocial behaviors and attitudes. On all other scales, the victims scored higher than the offenders, or no statistically significant differences were found between the groups (e.g., TRIN-r, L-r, BXD, RC7, SFD, NFC, JCP, ACT, MEC, AGG-r, and NEGE-r). The fact that no differences were found on some scales may suggest a profile overlap, wherein both samples showed similar behavioral problems, self-doubt, a sense of inefficacy, a history of conduct problems, and a variety of negative emotional experiences.

Considering that most participants had experienced traumatic events, we also investigated the influence of ACEs on the PAI and MMPI-2-RF scores. Regression analyses showed that the number of ACEs significantly predicted the PAI (e.g., NIM, PIM, SOM, ANX, ARD, DEP, MAN, PAR, SCZ, BOR, ANT, DRG, AGG, SUI, STR, NON, RXR, and WRM) and MMPI-2-RF scores (e.g., VRIN-r, TRIN-r, Fs, RBS, THD, BXD, RC1, RC7, RC8, RC9, MLS, SHY, DSF, MEC, PSYC-r, and DISC-r.). Taken together, these results are consistent with the literature that links ACEs to criminal behavior (e.g., Kahn et al. 2021), substance use disorders (Rhee et al. 2019), and psychopathology (e.g., Sheffler et al. 2020). The assumption that both victims and inmates have a history of victimization is coherent with the literature that states that ACEs are associated with perpetration and victimization of physical aggression (e.g., Nikulina et al. 2021). We can therefore hypothesize that, regardless of whether a person is a victim or a perpetrator, their childhood experiences tend to be similar when it comes to traumatic experiences.

To investigate the contribution of isolated ACEs in accurately differentiating between the victims and the offenders, we conducted discriminant analyses. Although both groups experienced ACEs, specific ACEs appeared to differentiate the victims from the offenders, particularly watching a parent being repeatedly hurt with or without a weapon, witnessing a parent being kicked or hit, and witnessing a parent being pulled, grabbed, bitten, or hit with objects. In general, these findings were consistent with the studies that showed that exposure to interparental violence or other types of household dysfunction is linked to criminal behaviors and criminal propensity (e.g., Basto-Pereira et al. 2022; Cascardi et al. 2020). However, although some studies agree that both victims and offenders share ACEs (e.g., physical, emotional, and sexual abuse, or household dysfunction), it is not clear if they strongly predict victimization or criminal behavior (e.g., Navarro et al. 2022; Zhu et al. 2023).

Limitations and Future Directions

Notwithstanding the relevance of the present study, some limitations need to be highlighted. This study resorted to a female sample of victims who were involved in IPV cases. Therefore, there is a clear gap when it comes to other types of victimization experiences, unrelated to IPV. Future studies should include a wider sample of victims, namely victims of other violent crimes, white-collar crimes, or non-violent crimes. Similarly, our inmate sample was limited to a male group of perpetrators who served sentences for violent crimes. Therefore, future studies should include perpetrators of non-violent crimes and assess the differences between types of perpetrators and types of victims. Similarly, while offenders were convicted of committing several types of offenses, the victim group was limited to IPV victimization, which could add confounding variables, which were not considered in the present study. Therefore, future research should make sure to only include IPV victims and offenders. We did not assess female inmates, whose PAI profiles may be significantly different from those of male inmates. The inclusion of these participants may be crucial to obtain a clearer picture of the personality function of victims and inmates. Finally, differences in the MMPI-2-RF and PAI scores may reflect differences due to the sex of participants, which should be considered in future studies. Future research should also aim to assess the impact of age and education on the MMPI-2-RF and PAI scores when dealing with victimized and offending populations.

5. Conclusions

The present study is an important step towards a better understanding of the role of certain variables (e.g., ACEs) in crime victimization and perpetration, helping researchers to comprehend the relative overlap between victims and offenders, providing relevant findings for forensic psychologists during the evaluation and decision-making, and supporting their considerations in reports.

Overall, the results confirmed past research on the personality of both offenders and victims, in that they provide a thorough comparison between the groups and reinforce the idea of an overlap in many personality traits, while simultaneously distinguishing these groups through other personalistic characteristics.

The PAI and the MMPI-2-RF are two objective measures of personality that have garnered international recognition, and which have been used in a wide array of contexts, including forensic assessments. The present study reinforced the applicability of these two instruments in offender and victim populations, as they were able to identify both profile overlap and different psychological traits between these two groups. Overall, our results show that the MMPI-2-RF appears to be more capable of differentiating between these populations than the PAI, whose ability to discriminate between groups was more modest. Similarly, when it comes to the impact of ACEs on psychological profiles, our results suggest that the MMPI-2-RF was superior, at least with regard to the regression models obtained. Nevertheless, both measures were sensitive to adverse childhood experiences, and most PAI scales were affected by ACEs, proving that this instrument is also sensitive to these types of experiences. Furthermore, differences were too small to draw any definite conclusions on which is the better assessment instrument overall.

From a practical perspective and taking into account that both assessment tools provide valuable and differentiated data on the functioning of their examinees, we support a simultaneous and complementary administration of these two measures, considering that they provide highly robust, precise, and comprehensive descriptions of forensic subjects. Nevertheless, professionals should be careful not to generalize findings and reduce IPV victims or offenders to results from a single measure of personality or psychopathology. Instead, a multi-source, multimethod approach should always be prioritized (Fokas and Brovko 2020).

The PAI and the MMPI-2-RF, as objective measures of personality and psychopathology, have shown to be robust, sensitive, and comprehensive instruments that are able to characterize both victims and offenders, delineate intervention programs, and understand the factors associated with the resistance to treatment (among others). This study contributes to validation efforts of the PAI, particularly with forensic samples of victims and offenders. Its forensic applicability, widely documented in the literature, was, therefore, confirmed in the present study, and its results and qualities were comparable to the ones obtained using the MMPI-2-RF.